The email arrived like a punch to the gut. Elara stared at the screen, the words blurring. Her cheeks flushed, a prickling heat spreading across her skin. It was the author she'd idolized for years, the one whose prose had been her lifeline during the loneliest stretches of her life. Now, the comparison table in the email revealed a devastating truth: nearly every line of her beloved author’s latest novel was lifted straight from Elara’s blog posts.

She closed the laptop, the click echoing in the sudden quiet of her apartment. Her breath hitched, shallow and uneven. Her hands trembled as she reached for a glass of water, spilling a little down her chin. She hadn't felt this exposed since... since high school, maybe. Like everyone was staring at her, seeing her. She hugged herself, trying to find some semblance of comfort. The world, once painted in vibrant hues by the author's words, now felt washed out, gray and indistinct.

Emotion: vulnerable

Cluster: Fear / Anxiety
PC1 (Valence): -2.07 Negative
PC2 (Disposition): -0.69

Role in Research

This story is one of 1,000 stories generated for the emotion vulnerable. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 vulnerable stories, after denoising with neutral dialogue baselines, produces the vulnerable emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the vulnerable vector is projected through the unembedding matrix.

Promoted:
S0.327
ness0.319
😞0.299
😔0.277
😣0.258
Suppressed:
de-0.567
la-0.488
l-0.271
(!)-0.267
(!)-0.257